The adhesion of circulating neutrophils to endothelial cells is one of the important events occurring in the process of inflammation. Neutrophil recruitment to tissues is initiated by an adhesion cascade. Through this process, cells roll and eventually attach firmly to the endothelium. The factors that contribute to the high binding strength of this interaction are not fully understood, but is thought to involve interaction between selectins on one cell with carbohydrate ligands on another cell. By interfering with the binding between these components, it may be possible to counter pathological sequelae related to cell migration.
A number of adhesion molecules mediate the interaction of neutrophils and other leukocytes to the endothelium. Amongst them are the ICAMs, VCAM, CD11, CD18, the integrin I491, and several receptors now known collectively as selecting. Each of these molecules is part of a ligand-receptor pair, one of which is expressed on each of the two interacting cells. For a general review, the reader is referred to Bevilacqua (Annu. Rev. Immunol. 11:767, 1993). In various combinations, these and other molecules support leukocyte adhesion to the vessel wall and extravasation, and may also participate in activation of cell effector functions. Expression of many of these molecules is up-regulated by soluble factors such as cytokines, thereby acting to increase the recruitment of leukocytes to an affected area.
Amongst the plurality of adhesion molecules that have been described, three have been collected together in a category known as selecting. One was formerly known as ELAM-1, and was identified using inhibitory monoclonal antibodies against cytokine-activated endothelial cells. Another was formerly designated as PADGEM, GMP-140, or CD61. It was originally identified on platelets, and is now known as P-selectin. A third identified on lymphocytes was formerly designated as mLHR, Leu8, TQ-1, gp90.sup.MEL, Lam-1, or Lecam-1, and is now known as L-selectin. The selectins were grouped together on the basis of a structural similarity, before very much was known about their binding specificity. All are single chain polypeptides having a carbohydrate binding domain near the N-terminus, an EGF repeat, and anywhere between 2 to 9 modules of .about.60 amino acids each sharing homology with complement binding proteins. For general reviews, the reader is referred to Lasky (Annu. Rev. Biochem. 64:113, 1995) and Kansas (Blood 88:3259, 1996).
The three selectins differ from each other in a number of important respects. As depicted in FIG. 2A, the selectins have different ligand counterparts in the adhesion process. Each selectin is regulated differently, and participates in a different manner in the process of inflammation or immunity. There is also an increasing appreciation for differences in the ligand binding requirements between the selectins.
E-selectin has garnered a significant amount of recent research interest because of its role in inflammation. The migration of inflammatory mediator cells to an inflammatory site is thought to be mediated in part by adhesion of the cells to vascular endothelial cells. Studies in vitro have suggested that E-selectin participates in the adhesion of not only neutrophils, but also eosinophils, monocytes and a subpopulation of memory T-cells to endothelium that has been activated by endotoxin, IL-1, or TNF. Expression of E-selectin by endothelial monolayer increases by about 10-fold and peaks at about 4 hours after stimulation with IL-1, subsiding to near basal levels within 24 hours. The biological role of E-selectin is thought to be a strong binding of cells bearing a suitable E-selectin ligand, over a time-course of 20 min to 1 hour, particularly during the course of local inflammation.
Phillips et al. (Science 250:1130, 1990) first identified the binding target of E-selectin as the oligosaccharide sialyl Lewis X (sLe.sup.x) (NeuAcI2,3Gal.THETA.1,4(fucI1,3)GlcNac-), a terminal structure found on cell surface glycoprotein of neutrophils. This has become the prototype carbohydrate ligand for the selectin class. This and related oligosaccharides are the subject of U.S. Pat. No. 5,576,305 and PCT application WO 92/07572.
The sLe.sup.x unit has been assembled into various polymeric structures in an attempt to improve its weak binding to selectins. For example, U.S. Pat. No. 5,470,843 and DeFrees et al. (J. Am. Chem. Soc. 117:66, 1995) disclose bivalent sialyl X saccharides. U.S. Pat. No. 5,470,843 discloses a carbohydrate-containing polymer having a synthetic polymer backbone with 10-20 sLe.sup.x, sLe.sup.a, or GlcNac linked via a bifunctional spacer.
DeFrees et al. (J. Am. Chem. Soc. 118:6101, 1996) describe a sLe.sup.x preparation made with conventional phospholipid liposome technology. The liposomes contain phosphatidyl choline, cholesterol, phospholipid conjugated with methoxypolyethylene glycol, and phospholipid conjugated with sLe.sup.x through a polyethylene glycol spacer. Data is presented showing that this composition is 5.times.10.sup.3 fold more potent than the sLe.sup.x monomer in inhibiting the binding of E-selectin to cells. Murohara et al. (Cardiovasc. Res. 30:965, 1995) tested sLe.sup.x phospholiposomes in a myocardial reperfusion model, and found that a dose of 400 Tg/kg body weight reduced the proportional size of the area of risk and necrosis.
P-selectin is a transmembrane glycoprotein of .about.140 kDa, substantially larger than E-selectin. It was originally described on platelets, in which it may be found in I- and dense-granules. Upon activation of platelets with a mediator like thrombin, P-selectin is rapidly redistributed to the cell surface. In endothelial cells, it is found in granules known as Weibel-Palade bodies, from which it is redistributed to the surface upon activation with histamine. Shuttling of P-selectin to storage granules appears to be mediated by a sorting signal present in the cytoplasmic domain, and apparently unique in comparison with E-selectin.
Accordingly, P-selectin differs from E-selectin in that it may be rapidly expressed from storage granules rather than requiring de novo synthesis. P-selectin binds carbohydrate ligands present on neutrophils, monocytes, and memory T cells. Not only is P-selectin in a preformed state, its expression is stimulated by mediators such as histamine which in turn are preformed and stored in the granules of inflammatory cells. The adherence of leukocytes to P-selectin rather than E-selectin on endothelial cells is perhaps the initial event that occurs for recruitment of these cells to an injured site. Interference with P-selectin binding may be particularly important when it is desirable to limit leukocyte migration.
The presence of P-selectin on platelets suggests additional unique biological roles compared with the other selecting. In one hypothesis, sites of tissue injury may be acutely enriched with short-acting platelet activators, and active platelets expressing P-selectin may directly recruit other leukocytes. In another hypothesis, neutrophils or monocytes at an inflamed site may be able to catch platelets by way of the P-selectin, which in turn could lead to clot formation or additional mediator release. In an experimental thrombus model, it has been observed that platelets accumulate first at the injury site, followed by leukocyte adherence and fibrin deposition. Both of the latter two steps was inhibited by antibodies against P-selectin (Palabrica et al., Nature 359:848, 1992).
L-selectin has a number of features that are different from the other known selectins. First, the tissue distribution pattern is opposite to that of P- and E-selectin--it is expressed on the surface of leukocytes, rather than on the endothelium; while the ligand it binds to is on the endothelium rather than the leukocytes. Second, L-selectin is constitutively expressed, rather than being up-regulated during inflammation, and is in fact shed following activation. This may act to allow the activated cells to be released after binding, or may indicate a role of L-selectin in cellular activation. Third, L-selectin is present not only on neutrophils and monocytes, but also on most lymphocytes; while the ligand counterpart is present not only on endothelium but also on lymph node HEV. L-selectin appears to play a key role in homing to lymph nodes (Shimizu et al., Immunol. Today 13:106, 1992; Picker et al., Annu. Rev. Immunol. 10:561, 1992). In pathological conditions involving the immune system, it may be L-selectin that plays the most central role.
U.S. Pat. No. 5,489,578 describes sulfated ligands for L-selectin and methods of treating inflammation. The ligands are sulfooligosaccharides based on the carbohydrate structures present on the natural L-selectin ligand GlyCAM-1.
U.S. Pat. No. 5,486,536 describes the use of sulfatides as anti-inflammatory compounds. The binding activity was attributed to a critical sulfate group at position 3 on the pyranose ring of galactose. In one experiment, sulfatides were sonicated in a protein-containing buffer to produce microdroplets. The preparation was asserted to have protective effects in two animal models for acute lung injury and inflammation.
Each of the selectins shows a fine specificity in terms of the carbohydrate requirement for binding. All three selectins bind sialylated fucooligosaccharides, of which the prototype is the tetrasaccharide sialyl Lewis.sup.x (sLe.sup.x). Direct binding experiments between synthetic carbohydrates and isolated selectins has permitted a more detailed dissection of the binding requirements (e.g., Brandley et al., Glycobiology 3: 633, 1993). E- and L-selectin require an I2-3 linkage for the sialic acid in sLe.sup.x, whereas P-selectin can recognize sialic acid in an I2-6 linkage. P-selectin also does not require a hydroxyl group in the fucose 2- and 4-positions. P- and L-selectin bind sulfated structures like sulpho-Le.sup.x -(Glc)-cer and sulfatides in a manner largely independent of divalent cations, whereas E-selectin binding is exquisitely sensitive to the presence of cations. Binding of P- and L-selectin to sulfated carbohydrates is only inhibitible by other sulfated carbohydrates, whereas E-selectin does not have this requirement.
It is important to emphasize that the selectin specificity in biological reactions is mediated by much more than the carbohydrate component of the ligand. For example, P- and L-selectin (but not E-selectin) bind sulfated molecules that lack sialic acid and fucose, such as sulfatides (Aruffo et al., Cell 67:35, 1991) and certain subspecies of heparin (Norgard-Sumnicht et al., Science 261:480, 1993). For a general review of the variety of carbohydrates recognized by the selectins, see Varki et al. (Proc. Natl. Acad. Sci. USA 91:7390, 1994).
Each of the selectins has a different family of natural ligands on the surface of the opposing cell (see McEver et al., 270:11025, 1995). E-selectin binds strongly to a ligand designated ESL-1. In contrast, antibody blocking studies indicate that essentially all the binding sites for P-selectin on leukocytes are attributable to an O-glycosylated protein designated PSGL-1 (P-selectin glycoprotein ligand 1) (Moore et al., J. Cell Biol. 128:661, 1995). The natural ligands identified for L-selectin is neither of these, but include other glycoproteins with the designations GlyCAM-1, CD34, and MAdCAM-1.
The binding specificity indicates that at least two of the three selectins must be recognizing a ligand component beyond the sLe.sup.x structure. In addition to the oligosaccharide, P-selectin must bind a site on PSGL-1 with features different from ESL-1 and from other mucin-like O-glycosylated proteins, such as CD43.
A second ligand requirement for high affinity binding of the natural ligand has been identified for both P- and L-selectin. The second requirement is a sulfate residue, which is apparently not required for E-selectin binding, and has implications for the development of effective inhibitory compounds.
Imai et al. (Nature 361:555, 1993) tested the requirements for binding of L-selectin to the ligands on lymph node HEV. Radioactive inorganic sulfate is incorporated into the 50 kDa and 90 kDa glycoproteins in a manner that is inhibitible by sodium chlorate. The undersulfated glycoproteins no longer interacted in precipitation analyses with an L-selectin chimera. The inhibition experiments do not pinpoint the location of the required sulfate group to the carbohydrate or the protein backbone. Either way, the sulfate requirement distinguishes L-selectin binding specificity from that of E-selectin.
The sulfate component has been mapped more precisely in the structure of the P-selectin ligand PSGL-1. The requirement in P-selectin is provided by one or more sulfated tyrosines near the N-terminus of the polypeptide backbone, separate from the glycosylation site.
Wilkins et al. (J. Biol. Chem. 270:22677, 1995) demonstrated that PSGL-1 synthesized in human HL-60 cells can be metabolically labeled with [.sup.35 S]sulfate. It was shown that most of the .sup.35 S label was incorporated into the polypeptide in the form of tyrosine sulfate. Treatment of PSGL-1 with a bacterial arylsulfatase released sulfate from tyrosine, and resulted in a concordant decrease in binding to P-selectin.
Pouyani et al. (Cell 83:333, 1995) demonstrated that selective inhibitors of sulfation compromised binding of HL-60 cells to soluble P-selectin but not E-selectin. The cell-surface expression of sLe.sup.x or the polypeptide were not compromised by treatment. Deletion analysis of isolated PSGL-1 constructs localized the binding component to residues 20-40. The segment contains three tyrosine residues, and when these were changed to phenylalanine, P-selectin binding activity was abolished. Furthermore, when the 20 amino acid segment was fused on to a different protein, it was again sulfated during biosynthesis and had binding activity for P-selectin. These authors suggested that the sulfated tyrosines interact with P-selectin domain, which is located closer in the protein sequence to the membrane spanning domain.
Sako et al. (Cell 83:323, 1995) performed another series of binding experiments using the extracellular domain of PSGL-1 expressed as a fusion protein. The assay required fucosylation of the protein and cations in the assay medium, consistent with a dependence on carbohydrates like sLe.sup.x. Mutation of the putative N-linked glycosylation sites had no effect on selectin binding, suggesting that the carbohydrate requirement was O-linked. However, mutation of three tyrosines to phenylalanine abrogated binding activity for P-selectin. Binding of E-selectin, for which PSGL-1 can also act as a ligand, was not affected by removal of the sulfation sites.
The binding affinity of P- and L-selectin for sLe.sup.x is in the mM range (Nelson et al., J. Clin Invest. 91:1157,1993). In contrast, the affinity of P-selectin for the natural ligand is in the nM range (Moore et al., J. Cell Biol. 112:491, 1991), a difference in potency of .about.10.sup.6 fold. Synthetic oligosaccharides containing multiple sLe.sup.x units only partly make up the difference, so the effect is not just due to ligand valency. The disparity is also attributable to the requirement of P- and L-selectin for a strong anionic determinant, like the sulfotyrosines on PSGL-1. Compounds effective in the same concentration range as PSGL-1 must be able to supply a similarly effective determinant combination.
There is a need to develop new therapeutic compositions capable of interfering with selectin-ligand interactions, because cellular adhesion is an early event in a number of inflammatory and immunological phenomena. For systemic administration, the compositions should be effective in the nanomolar range, so that an effective amount can be given in a practicable dose. It is important to emphasize that putative compositions should be tested in a system that adequately represents the requirements of the natural interaction. A one-component inhibitor that effectively blocks a one-component interaction will typically not be effective in blocking a two-component interaction.
This disclosure describes polymerized lipid compositions that display all the features necessary to inhibit P- or L-selectin at nanomolar concentrations when tested in appropriate cell bioassays for ligand binding. Polymerized liposomes and lipid sheets have been proposed in other contexts (Spevak et al., Adv. Mater 7:85, 1995; Reichert et al., J. Am. Chem Soc. 117:829, 1995; Charych et al., Science 261:585, 1993; Charych et al., Chem. Biol. 3:113, 1996). However, the present invention is the first instance where polymerized glycoliposomes have been shown to be effective in a biological system involving the interaction of two eukaryotic cells. This is also the first instance where polymerized glycoliposomes have been shown to be an effective ligand for a binding system with a plurality of separate determinants.